Handle leading 0s in group refs, fixes #88

CHANGELOG.md

@@ -7,6 +7,12 @@ and this project adheres to [Semantic Versioning](https://semver.org/spec/v2.0.0
 
 ## [Unreleased]
 
+### Fixed
+
+- fixed scanner errors for insignificant leading zeros in numerical group refs
+  * e.g. `(a)\k<01>`, `(a)\g<-01>`, `(a)?(?(01)b|c)`
+  * thanks to [Markus Schirp](https://github.com/mbj) for the report
+
 ## [2.8.2] - 2023-10-10 - Janosch Müller
 
 ### Fixed

lib/regexp_parser/scanner/scanner.rl

@@ -364,6 +364,7 @@
   conditional_expression := |*
     group_lookup . ')' {
       text = copy(data, ts, te-1)
+      text =~ /[^0]/ or raise ValidationError.for(:backref, 'condition', 'invalid ref ID')
       emit(:conditional, :condition, text)
       emit(:conditional, :condition_close, ')')
     };
@@ -541,13 +542,13 @@
       case text = copy(data, ts, te)
       when /^\\k(.)[^0-9\-][^+\-]*['>]$/
         emit(:backref, $1 == '<' ? :name_ref_ab : :name_ref_sq, text)
-      when /^\\k(.)[1-9]\d*['>]$/
+      when /^\\k(.)0*[1-9]\d*['>]$/
         emit(:backref, $1 == '<' ? :number_ref_ab : :number_ref_sq, text)
-      when /^\\k(.)-[1-9]\d*['>]$/
+      when /^\\k(.)-0*[1-9]\d*['>]$/
         emit(:backref, $1 == '<' ? :number_rel_ref_ab : :number_rel_ref_sq, text)
       when /^\\k(.)[^0-9\-].*[+\-]\d+['>]$/
         emit(:backref, $1 == '<' ? :name_recursion_ref_ab : :name_recursion_ref_sq, text)
-      when /^\\k(.)-?[1-9]\d*[+\-]\d+['>]$/
+      when /^\\k(.)-?0*[1-9]\d*[+\-]\d+['>]$/
         emit(:backref, $1 == '<' ? :number_recursion_ref_ab : :number_recursion_ref_sq, text)
       else
         raise ValidationError.for(:backref, 'backreference', 'invalid ref ID')
@@ -560,9 +561,9 @@
       case text = copy(data, ts, te)
       when /^\\g(.)[^0-9+\-].*['>]$/
         emit(:backref, $1 == '<' ? :name_call_ab : :name_call_sq, text)
-      when /^\\g(.)\d+['>]$/
+      when /^\\g(.)(?:0|0*[1-9]\d*)['>]$/
         emit(:backref, $1 == '<' ? :number_call_ab : :number_call_sq, text)
-      when /^\\g(.)[+-]\d+/
+      when /^\\g(.)[+-]0*[1-9]\d*/
         emit(:backref, $1 == '<' ? :number_rel_call_ab : :number_rel_call_sq, text)
       else
         raise ValidationError.for(:backref, 'subexpression call', 'invalid ref ID')

spec/parser/conditionals_spec.rb

@@ -66,4 +66,8 @@
   include_examples 'parse', /(?<A>a)(?(<A>)T|)/,
     [1]    => [Conditional::Expression, count: 3, to_s: '(?(<A>)T|)'],
     [1, 2] => [Conditional::Branch, to_s: '', ts: 16]
+
+  # test insignificant leading zeros in the condition's group number ref
+  include_examples 'parse', /(a)(?(001)T)/,
+    [1, 0] => [Conditional::Condition, to_s: '(001)', reference: 1]
 end

spec/parser/refcalls_spec.rb

@@ -6,12 +6,16 @@
   include_examples 'parse', /(?<X>abc)\k<X>/, 1 => [Backreference::Name, name: 'X', reference: 'X']
   include_examples 'parse', /(?<X>abc)\k'X'/, 1 => [Backreference::Name, name: 'X', reference: 'X']
   include_examples 'parse', /(abc)\k<1>/,     1 => [Backreference::Number, number: 1, reference: 1]
+  include_examples 'parse', /(abc)\k<001>/,   1 => [Backreference::Number, number: 1, reference: 1]
   include_examples 'parse', /(abc)\k<-1>/,    1 => [Backreference::NumberRelative, number: -1, reference: 1]
   include_examples 'parse', /(abc)\k'-1'/,    1 => [Backreference::NumberRelative, number: -1, reference: 1]
+  include_examples 'parse', /(abc)\k'-001'/,  1 => [Backreference::NumberRelative, number: -1, reference: 1]
   include_examples 'parse', /(?<X>abc)\g<X>/, 1 => [Backreference::NameCall, reference: 'X']
   include_examples 'parse', /(abc)\g<1>/,     1 => [Backreference::NumberCall, reference: 1]
+  include_examples 'parse', '(abc)\g<001>',   1 => [Backreference::NumberCall, reference: 1]
   include_examples 'parse', '\g<0>',          0 => [Backreference::NumberCall, reference: 0]
   include_examples 'parse', /(abc)\g<-1>/,    1 => [Backreference::NumberCallRelative, reference: 1]
+  include_examples 'parse', /(abc)\g<-001>/,  1 => [Backreference::NumberCallRelative, reference: 1]
   include_examples 'parse', /\g<+1>(abc)/,    0 => [Backreference::NumberCallRelative, reference: 1]
 
   include_examples 'parse', /(?<X>abc)\k<X-0>/,

spec/scanner/conditionals_spec.rb

@@ -14,6 +14,7 @@
   include_examples 'scan', /(a)(?(1)TRUE|)11/,    9  => [:conditional,  :close,           ')',   13, 14]
   include_examples 'scan', /(?<N>A)(?(<N>)T|F)1/, 5  => [:conditional,  :condition,       '<N>', 10, 13]
   include_examples 'scan', /(?'N'A)(?('N')T|F)2/, 5  => [:conditional,  :condition,       "'N'", 10, 13]
+  include_examples 'scan', /(a)(?(001)T)/,        5  => [:conditional,  :condition,       '001', 6,  9]
 
   include_examples 'scan', /(a(b(c)))(?(1)(?(2)d|(?(3)e|f))|(?(2)(?(1)g|h)))/,
     0  => [:group,        :capture,         '(',   0,  1],

spec/scanner/errors_spec.rb

@@ -66,8 +66,15 @@
   include_examples 'scan error', RS::InvalidBackrefError, 'invalid backref', '\k\'\''
   include_examples 'scan error', RS::InvalidBackrefError, 'invalid backref', '\k<0>'
   include_examples 'scan error', RS::InvalidBackrefError, 'invalid backref', '\k\'0\''
+  include_examples 'scan error', RS::InvalidBackrefError, 'invalid backref', '\k<-0>'
+  include_examples 'scan error', RS::InvalidBackrefError, 'invalid backref', '\k<000>'
+  include_examples 'scan error', RS::InvalidBackrefError, 'invalid backref', '\k<-000>'
   include_examples 'scan error', RS::InvalidBackrefError, 'invalid refcall', '\g<>'
   include_examples 'scan error', RS::InvalidBackrefError, 'invalid refcall', '\g\'\''
+  include_examples 'scan error', RS::InvalidBackrefError, 'invalid refcall', '\g<000>'
+  include_examples 'scan error', RS::InvalidBackrefError, 'invalid refcall', '\g<-000>'
+  include_examples 'scan error', RS::InvalidBackrefError, 'invalid condition', '(a)(?(0)b)'
+  include_examples 'scan error', RS::InvalidBackrefError, 'invalid condition', '(a)(?(000)b)'
   include_examples 'scan error', RS::UnknownUnicodePropertyError, 'unknown property', '\p{foobar}'
   include_examples 'scan error', RS::UnknownPosixClassError, 'unknown POSIX class [::]', '[[::]]'
   include_examples 'scan error', RS::UnknownPosixClassError, 'unknown POSIX class [:^:]', '[[:^:]]'

spec/scanner/refcalls_spec.rb

@@ -20,9 +20,11 @@
 
   include_examples 'scan', '(abc)\k<1>',        3 => [:backref, :number_ref_ab,           '\k<1>',      5, 10]
   include_examples 'scan', "(abc)\\k'1'",       3 => [:backref, :number_ref_sq,           "\\k'1'",     5, 10]
+  include_examples 'scan', "(abc)\\k'001'",     3 => [:backref, :number_ref_sq,           "\\k'001'",   5, 12]
 
   include_examples 'scan', '(abc)\k<-1>',       3 => [:backref, :number_rel_ref_ab,       '\k<-1>',     5, 11]
   include_examples 'scan', "(abc)\\k'-1'",      3 => [:backref, :number_rel_ref_sq,       "\\k'-1'",    5, 11]
+  include_examples 'scan', '(abc)\k<-001>',     3 => [:backref, :number_rel_ref_ab,       '\k<-001>',   5, 13]
 
   # Sub-expression invocation, named, numbered, and relative
   include_examples 'scan', '(?<X>abc)\g<X>',    3 => [:backref, :name_call_ab,            '\g<X>',      9, 14]
@@ -33,12 +35,14 @@
 
   include_examples 'scan', '(abc)\g<1>',        3 => [:backref, :number_call_ab,          '\g<1>',      5, 10]
   include_examples 'scan', "(abc)\\g'1'",       3 => [:backref, :number_call_sq,          "\\g'1'",     5, 10]
+  include_examples 'scan', '(abc)\g<001>',      3 => [:backref, :number_call_ab,          '\g<001>',    5, 12]
 
   include_examples 'scan', 'a(b|\g<0>)',        4 => [:backref, :number_call_ab,          '\g<0>',      4, 9]
   include_examples 'scan', "a(b|\\g'0')",       4 => [:backref, :number_call_sq,          "\\g'0'",     4, 9]
 
   include_examples 'scan', '(abc)\g<-1>',       3 => [:backref, :number_rel_call_ab,      '\g<-1>',     5, 11]
   include_examples 'scan', "(abc)\\g'-1'",      3 => [:backref, :number_rel_call_sq,      "\\g'-1'",    5, 11]
+  include_examples 'scan', '(abc)\g<-001>',     3 => [:backref, :number_rel_call_ab,      '\g<-001>',   5, 13]
 
   include_examples 'scan', '\g<+1>(abc)',       0 => [:backref, :number_rel_call_ab,      '\g<+1>',     0, 6]
   include_examples 'scan', "\\g'+1'(abc)",      0 => [:backref, :number_rel_call_sq,      "\\g'+1'",    0, 6]